Computer Vision-based Sorting And Recycling System For Waste Plastic Bottles

Plastics are widely used in all walks of life,on account of their various fine properties,including light weight,chemical stability,anti-corrosion,impact resistance,great stainability and high processing cost,etc.Alongside that,plastic bottles are the dominant plastic products.Therefore,the recycling of waste plastic bottles is of great significance.Sorting plastic bottles on the basis of their colors is one of the core technologies of recycling plastic bottles.At present,sorting plastic bottles in china mainly relies on manual work,which has a heavy workload and low efficiency.It is not conducive to the large-scale recycling of waste plastic bottles.So more research on bottle recycling is essential.This essay will analyze the sorting process and define the technological requirements of visual sorting according to the real production demand of waste plastic bottles sorting.In order to solve the problem of overlapping plastic bottles,the image outline feature extraction and Euclidean distance classification algorithm based on the Zernike moment are proposed to effectively recognize the overlapping plastic bottles and avoid the interference from overlapping plastic bottles to the sorting of old plastic bottles by color.Aiming at avoiding the interference from bottle caps,a method of image area weighting is proposed to reduce the interference from bottle caps to subsequent image classification algorithm.To deal with the residual labels on plastic bottles after it has been removed,a recognition algorithm is proposed to realize the recognition of plastic bottles based on feature weighted image texture and feature extraction.In order to solve the problem of sorting plastic bottles according to the color,a multi-channel image color feature extraction method is proposed.The color feature description vector of the image is composed of the color moments of H,S and V channels of the image.The kernel parameters and penalty parameters of the support vector machine are optimized by genetic algorithm to obtain the optimal support vector machine model.The optimized support vector machine model can be used to classify and recognize the image features.That could effectively improve the accuracy of image recognition.Finally,the effectiveness and feasibility of the algorithm are verified by experiments.